March 2004

The Socioeconomic Status of Black Males: The Increasing Importance of Incarceration

Steven Raphael

Goldman School of Public Policy University of California, Berkeley

E-mail: raphael@socrates.berkeley.edu

For Geno, my mentor and friend.

Abstract This paper assesses the increasing importance of incarceration in determining the average socioeconomic status of black males in the United States. I document national trends in the proportion of black males that are either currently institutionalized or who have served previous prison time. The paper also documents the extent to which serving time interrupts the potential early work careers of young offenders and reviews recent research on employer sentiment regarding ex-offenders and the likely stigma effects of prior incarceration. Finally, I assess whether increasing incarceration rates provide a possible explanation for the drastic declines in employment rates observed among non-institutionalized black males. Using data from the U.S. Census, I test for a correlation between the proportion of non-institutionalized men in a given age-race-education group that are employed and the proportion of all men in this grouping that are institutionalized. The proportion institutionalized has a strong negative effect on the proportion of the non-institutionalized that are employed. The relationship is strong enough to explain one-third to one-half of the relative decline in black male employment rates.

1. Introduction

Over the past three decades, the average socioeconomic status of African-American males

has deteriorated, absolutely and relative to men from other racial and ethnic groups. Despite

gains in relative earnings immediately following passage of the Civil Rights Act, the relative

earnings of black men have stagnated since the mid-1970s (Freeman 1992). In addition,

employment rates among non-institutionalized black men have declined markedly, with

pronounced declines for the relatively less educated (Holzer and Offner 2003).

Concurrent with these adverse labor market trends is a phenomenal increase in the proportion

of black men involved in one form or another with the criminal justice system. Between 1970

and 2000, the proportion of working age black males that are institutionalized increased from 3

to 8 percent. Over the same time period, the proportion of black males that have ever served a

prison sentence increased from approximately 7 to 17 percent (Bonczar 2003). For both

measures, these increases were considerably larger for certain subgroups of the black male

population, such as the relatively less educated and the young.

The increasing proportion of black men that are either currently incarcerated or have been

incarcerated in the past is likely to suppress their relative socioeconomic status. Having served a

prison sentence worsens one’s labor market prospects for a variety of reasons, and thus

aggravates stubborn racial differences in employment and pay. In addition, the interruption of a

prison sentence and the consequent lengthy absence of African-American males from their

spouses, partners, and children is likely to hamper family formation and hasten the dissolution of

existing, if not stable, family units. Given the high fraction of black men with felonious criminal

history records, these collateral consequences of incarceration are quickly becoming an

increasingly important source of racial inequality in the United States. In fact, one might argue

2

that in light of the potentially permanent consequences of an incarceration spell, the high

incarceration rate among black males is perhaps one of the chief barriers to their socioeconomic

progress.

This paper assesses the increasing importance of incarceration in determining the average

socioeconomic status of black males in the United States. I document national trends in the

proportion of black males that are institutionalized and how this proportion varies by age and

level of educational attainment. Using several decades of data from the U.S. Census of

Population and Housing, I demonstrate that the proportion of black men institutionalized has

increased markedly, while the comparable proportion of white men has not. I observe similar

trends for the fraction of men with current or previous prison experience. Combined, these

analyses demonstrate the high proportions of certain sub-groups of black men that are

incarcerated and the near universality of prior prison experience for some.

Next, I combine an analysis of California administrative data with a review of recent research

on employer demand for ex-offenders to explore the long-term employment consequences of

serving time. The California data reveal that many young offenders cycle in and out of prison

during a portion of the lifecycle when earnings tend to increase the most among the non-

institutionalized. Such cycling is likely to decrease the degree of attachment to the legitimate

labor market and impede the accumulation of valuable employment experience. In addition,

several recent surveys of employers reveal a strong reluctance to hire former prison inmates.

Combined, these findings indicate that serving time may permanently alter the age-earnings

profiles of prison inmates.

Finally, I assess whether increasing incarceration rates provide a possible explanation for the

drastic declines in employment rates observed among non-institutionalized black males. Using

3

data from the U.S. Census, I test for a correlation between the proportion of non-institutionalized

men in a given age-race-education group that are employed and the proportion of all men in this

grouping that are institutionalized. The proportion institutionalized has a strong negative effect

on the proportion of the non-institutionalized that are employed. The relationship is strong

enough to explain nearly half of the relative decline in employment rates of black men.

2. Changes in Incarceration Rates: 1970 to 2000

This section documents incarceration trends over the past three decades. I focus on two

measures of incarceration: the proportion of men institutionalized at a given point in time and the

proportion of men that are either currently incarcerated or have served time at some point in the

past. As we will see, the large fraction of currently incarcerated black men suggests that a much

larger fraction of this population is in a non-productive status than the traditional focus on the

employment rates of the non-institutionalized would suggest. In addition, the increasing

proportion of black men with previous prison experience indicates that many non-

institutionalized blacks face the employment barriers specific to ex-offenders.

Documenting Trends in Institutionalization from the U.S. Census

The decennial Census of Population and Housing enumerates both the institutionalized as

well as the non-institutionalized population. The Public Use Microdata Samples (PUMS) for

each census includes a flag for the institutionalized as well as micro-level information on age,

education, race and all other information available for non-institutionalized long-form

respondents. Within the institutionalized population, one can separately identify individuals

residing in non-military institutions. This category includes inmates of federal and state prisons,

local jail inmates, residents of inpatient mental hospitals, and residents of other non-aged

4

institutions. I use residence in a non-military institution as the principal indicator of

incarceration.1

To gauge the validity of using the census data in this manner, Figure 1 compares

estimates of the institutionalized population from the census to estimates of the incarcerated

populations from other sources. The figure presents a comparison of the number of

institutionalized adult black, white, and Hispanic males from the 2000 census to counts of the

number of prison and jail inmates at midyear 2001 calculated by the U.S. Bureau of Justice

Statistics (BJS) (Beck, Karberg, and Harrison 2002).2 Given the inclusive nature of the census

definition of institutionalization, the census estimates are slightly larger than the BJS numbers

for all groups. Nonetheless, the two sets of estimates correspond quite closely and the

differences are small.

Table 1 documents employment and incarceration trends by race and educational

attainment using data from the 1970, 1980, 1990, and 2000 one percent PUMS. The table

presents the proportion of non-Hispanic black and white males 18 to 65 years of age that are

employed, that are not working yet not institutionalized, that are in the armed forces, and that are

institutionalized. For all black men, the proportion employed declines markedly over this 30-

year period, from 0.73 in 1970 to 0.57 in the 2000. This decline occurs within all education

groups, although the drop is largest for black high school dropouts (from 0.71 to 0.34).

Employment rates decline slightly for white males overall, and decline substantially for white

high school dropouts. However, these changes are small in comparison to those observed for

blacks.

1 See Butcher and Piehl (1998) for an analysis of incarceration among immigrant men that also uses the group quarter variable to identify the incarcerated. 2 The BJS population estimates come from custody counts form the National Prisoner Statistics database and the 2001 Annual Survey of Jails. Thus, the census data and the BJS data come from entirely different sources.

5

Over the thirty year period, the proportion of black men that are institutionalized

increases considerably, especially for less educated black men. For all black males, the

proportion institutionalized increases nearly three-fold from 0.03 in 1970 to 0.08 in 2000. For

black high school dropouts, the institutionalization rate increases nearly five-fold. At the end of

the century, roughly one fifth of black men with less than a high school degree are

institutionalized. There is no increase in institutionalization among black males with at least a

college degree. Among whites, changes in institutionalization rates, overall and within

educational groups, are small by comparison.

Table 2 presents similar tabulations by age. For black men, the proportion

institutionalized increases within every age group, with the most pronounced increases for the

young. In 2000, roughly 11 percent of black men between 18 and 40 are institutionalized.

Again, while there are slight increases in the proportion institutionalized among young white

men, the changes are small in comparison to what we observe among African-Americans.

Tables 1 and 2 indicate that both age and educational attainment are strong predictors of

current incarceration. Table 3 explores the interaction between these two dimensions for black

men only. The table presents comparable tabulations for the subset of relatively young (under

40) and relatively less educated (dropouts and high school graduates) black men. For young high

school dropouts, the declines in the proportion employed are considerably more drastic than the

declines in employment for black male high school dropouts overall (presented in Table 1). For

dropouts between 18 and 25, the employment rate declines from 0.50 to 0.27. For those 26 to

30, the proportion employed declines from 0.76 to 0.30, while for 31 to 40 year olds,

employment rates decline from 0.81 to 0.35.

6

Similarly, increases in the proportions institutionalized are much larger than those

observed for dropouts overall. For dropouts between 18 and 25, the institutionalization rate

increases from 8 percent to 23 percent. For those between 26 and 40, the institutionalization

rates increases from approximately 5 percent to 30 percent. For all dropouts less than 40 years of

age, the institutionalized population is only slightly smaller than the population of employed men

from this demographic group. For black dropouts between 26 and 30, there are actually more

institutionalized than employed. Comparable, although somewhat muted, patterns are observed

for black high school graduates.

One factor that qualifies the figures presented in Tables 1 through 3 concerns the fact that

African-American men have historically been undercounted in the Census. While the 2000

enumeration made a concerted effort to improve coverage (and indeed, measured undercounting

was considerably lower than that observed for the 1990 census) disparities remain, with black

men again the most likely to be missed. The Accuracy and Coverage Evaluation (A.C.E.) results

released by Census Bureau (U.S. Census Bureau 2003) estimates that roughly 7 percent of non-

institutionalized black men were missed in the 2000 census. By age, the net undercount of non-

institutionalized black men is 6 percent for those 18 to 29, 10 percent for those 30 to 49, and 4

percent for those 50+.

To assess the effect of this undercount on my estimates of black male incarceration rates,

I calculate the following adjustment for the 2000 figures. Let Institution be the unadjusted

estimates of the proportion institutionalized, Non-institution be the comparable unadjusted

estimate for the household population, and Coverage be the proportion of the household

population measured in the census (one minus the net undercount rate). An adjusted

institutionalization rate can be calculated by the equation

7

./

*CoveragenInstitutioNonnInstitutio

nInstitutionInstitutio−+

=

For all black men, this adjustment decreases the estimate of the proportion institutionalized from

0.08 to 0.075. By age, the undercount adjustment reduced the proportion institutionalized from

0.11 to 0.105 for those 18 to 25, from 0.12 to 0.114 for those 26 to 30, from 0.11 to 0.10, for

those 31 to 40, from 0.06 to 0.54 for those 41 to 50, and from 0.03 to 0.029 for those 50 and

over. Thus, accounting for the undercount in the non-institutional population does not

appreciably affect the reported patterns.

The A.C.E. does not produce net undercount estimates by race, age, and level of

educational attainment. Thus it is difficult to present adjusted estimates for the results in Table 3

where we observe the starkest patterns. However, we can assume an arbitrarily high under-count

rate for this population and provide some lower-bound results. For example, if we assume that

one-fifth of black male high school dropouts are missed by the census (an undercount rate that is

twice the highest of the age-specific rates for black males), the estimate of the proportion

incarcerated among black high school dropouts declines from 0.23 to 0.19 for those 18 to 25,

from 0.34 to 0.29 for those 26 to 30, and from 0.28 to 0.24 for those 31 to 40. While these are

notable declines, these lower-bound estimates of the proportion incarcerated are still quite high.

Thus, institutionalization rates for black men have increased considerably since 1970.

These increases have been largest for the young and the relatively less educated. Interacting

these two dimensions reveals that the most dramatic increases in incarceration occurred among

black males with less than a high school education between the ages of 25 and 40.

8

Estimating the proportion with prior prison experience

While a sizable minority of black men is currently incarcerated, the fraction of this

population that has ever served time in state or federal prison is certainly larger. Turnover rates

in state and federal prisons are high and the median sentences fairly short. For example, the

median sentence for new prison admissions in the United States in 1999 was roughly three years

for the maximum sentence and 13 months for the minimum sentences. Moreover, many inmates

will serve considerably less time than their maximum sentences.

Consistent with the high degree of turnover among prisoners, the majority of inmates will

eventually be released. In 1997, 65 percent of surveyed inmates in state and federal prison

indicated that they had a definite release date. An additional 32 percent indicated that they

anticipated eventually leaving prison. Of the 97 percent of inmates that anticipated eventually

leaving, nearly 60 percent reported that they would be released within the next three calendar

years (Raphael and Stoll 2004). In conjunction, these figures suggest that the large increases in

the prison populations occurring over the last few decades have certainly left in their wake a

much larger population of former prisoners.

Gauging the population of former prison inmates is difficult due to the fact that none of

the major household surveys in the United States ask respondents whether they have served time.

Thus, estimating the size of this population requires indirect methods. The BJS estimates the

number of former inmates by combining population data, birth cohort estimates of the likelihood

of entering prison for the first time at each age (often separately by race and gender), and cohort

and age-specific mortality rates (Bonczar 2003).3 Based on this methodology, the BJS estimates

that in addition to the 1.3 million current inmates in 2001, an additional 4.3 million non-

9

institutionalized persons had served a prison term in the past. Combined, current and former

prison inmates account for 4.9 percent of the adult male population in 2001.

Of course, there are large differences by race and ethnicity. The same set of estimates

indicate that 2.6 percent of non-Hispanic white males, 16.6 percent of non-Hispanic black males,

and 7.7 percent of Hispanic males have served prison time (figures that are roughly double the

institutionalization rates listed in Table 1). The comparable figures for whites, blacks, and

Hispanics for 1974 were 1.4, 8.7, and 2.3 percent, respectively.

The BJS also uses this methodology to calculate lifetime probabilities of entering either

the state or federal prison system. Given that the risk of incarceration has increased over the past

three decades, lifetime probabilities should exceed the current proportion of a specific population

that is either currently incarcerated or formerly incarcerated.4 Figure 2 presents these estimates

for 1974 and 2001. For whites, the lifetime likelihood of going to prison for men born in 1974 is

estimated to be 2.2 percent. For those born in 2001, the risk increases to 5.9 percent. For black

males, this likelihood increases from 13.2 to 32.2 percent, while for Hispanics the likelihood

increases from 4 to 17.2 percent.

The analysis of institutionalization rates revealed large differences within racial groups

between less educated and more educated men and between groups of men stratified by age.

While the BJS does provide race-specific estimates of the proportion that has ever served time by

age, there are no estimates of how this proportion varies by level of educational attainment.

Moreover, the results presented above indicate that education is a stronger predictor of current

3 The likelihood of entering prison is estimated from annual surveys of recent prison admissions while mortality rates are based on mortality rates for the entire population adjusted upwards by a fixed factor to account for observed average differences in mortality rates between ex-offenders and the general population. 4 This is due to the fact that earlier cohorts faced lower risks of incarceration during the high-criminal-activity portion of their lifecycle.

10

incarceration than is age, and thus, education is also likely to be more strongly associated with

ever having served time.

I am able to fill this gap to a certain degree with administrative prison data from

California. Using administrative records on all prison terms served during the 1990s in a

California state prison, I first calculate an unduplicated count of prisoners entering the system

during the 1990s, by race and by how old each prisoner would be in the year 2000.5 I then use

the 1997 Survey of Inmates in State and Federal Correction Facilities to estimate the distribution

of inmates across age-education cells within racial and ethnic groups. These distribution

estimates are then used to allocate the number of unduplicated prisoners within each age-race

cell across educational attainment groups.6 Dividing these counts by the estimated 2000

California population (institutional plus non-institutional) within each age-race-educational

attainment cell yields estimates of the proportion of males in each cell serving a prison term

during the 1990s.

Table 4 presents these results. The first column presents national estimates of the

proportion ever serving time by race/ethnicity and age from the BJS. The second column

presents comparable estimates of the proportion serving time in California. The final four

columns present estimates by level of educational attainment that allot prisoner within race-age

cells across education groups according to the estimated educational distributions of inmates

during the late 1990s.

The tabulations by age indicate that the California estimates and the BJS estimates are

fairly similar for males between the ages of 18 and 54. For older males, the California estimates

5 Each record contains information on an internal California Department of Corrections id number that can be used to uniquely identify inmates. Thus, the administrative records can be purged of inmates that serve multiple prison spells. See Raphael and Weimer (2003) for a complete description of this administrative data set.

11

indicate a smaller proportion ever having served time. This is sensible considering that the

California administrative records only cover the 1990s, and that former prisoners over 54 in the

year 2000 are likely to have served time prior to the 1990s. Both sets of estimates indicate that

the proportion ever having served time increases with age through the late 30 and early 40s and

then declines. Black men between 25 and 44 have the highest rates of current or previous

incarceration (roughly one-fifth of this group using both the California and BJS estimates).

The estimates by race, age, and education reveal dramatic differences. For black high

school dropouts between the ages of 25 and 44, the number of unduplicated prisoners serving

time during the previous decade exceeds census population counts (--i.e., the ratio is greater than

one).7 For black high school dropouts between 45 and 54, 90 percent are estimated to have

served a prison term during the past decade. These figures suggest that for black high school

dropouts, serving time in prison is practically a certainty. The proportion of blacks with prison

time in the past decade is considerably lower for those with higher levels of educational

attainment, although the figures for black high school graduates are still quite high (between 0.12

and 0.16). By contract, the comparable fractions of whites as well as Latinos with prison time in

the previous 10 years are smaller for all comparisons.

6 The prisoner survey estimates of the joint age-education-race density is needed due to the fact that the California administrative records do not contain information on educational attainment. 7 To be sure, this does not mean that more than 100 percent of black men in this cell have served time in the past 10 years. There are a number of factors that are likely to bias upwards the count of unduplicated prisoners relative to the 2000 population. First, I calculated prisoner counts by age in 2000 without taking into account neither the likely mortality of many of the inmates serving time during the 1990s nor the likelihood that many of these inmates may have moved to another state after being released. In addition, a prisoner may be assigned additional internal CDC prisoner identification numbers for subsequent prison terms, thus artificially inflating the number of unduplicated spells. This however, is unlikely to be a substantial source of bias since tabulation based on prisoner SSN’s yield quite similar counts to the tabulations based on CDC identification codes. Finally, an undercount of black males in the census will suppress the denominator of this ratio below it’s actual level and inflate the rates reported above.

12

The effect of changing incarceration rates on estimates of black-white employment and earnings

trends

To summarize the patterns documented thus far, the current incarceration rates of black

men as well as the proportion of black men with prior prison experience have increased

considerably over the past three decades. These increases have been most pronounced for black

men who have not graduated from high school and who are between the ages of 25 and 40.

Moreover, while we observe increases in these rates for white men, the changes are minuscule by

comparison.

One interesting implication of these trends concerns the fact that traditional gauges of

black-white inequality, such as relative wages, unemployment rates, or employment-to-

population ratios are likely to suffer increasingly from selection bias. For example, the monthly

unemployment rate estimated by race and ethnicity by the Bureau of Labor Statistics and

reported widely in the press is calculated using data from the monthly Current Population Survey

(CPS). The sampling frame of the CPS is based on housing units, and thus pertains to the non-

institutionalized population. To the extent that incarcerated felons are highly likely to be

unemployed when not institutionalized, the increasing incarceration rate of blacks suggests that

monthly black unemployment rates are being calculated with an increasingly select sample.

Moreover, given the generally higher unemployment rate for blacks relative to other racial and

ethnic groups, estimates of the overall unemployment rate may be similarly biased downwards

relative to the counterfactual with lower incarceration rates.8 In general, use of non-

institutionalized samples to estimate labor market aggregates is likely to bias research findings

towards a greater degree of racial convergence.

13

Perhaps the most glaring example of selection bias is observed when comparing the

relative employment-to-population ratios of blacks and whites. Western and Petit (2000)

demonstrate that by adding the incarcerated to the denominator of the employment-to-population

ratio, white-black employment rate differentials increase considerably, especially for the

relatively young and the relatively less educated. To illustrate this point, Figures 3 presents

estimates of the difference in employment-to-population ratios between white and black men

using two alternative base populations: the non-institutionalized and the non-institutionalized

plus the institutionalized (labeled “All” on the graphs).9 The difference between the two sets of

estimates increase in each decade: in 1970, including the institutionalized increases the white-

black difference in employment rates by one percentage point, in 1980 by two, by three points in

1990, and by five percentage points in 2000. In 2000, the white-black employment rate

differential is 30 percent larger when the institutionalized are accounted for.

Table 5 presents comparable estimates by level of educational attainment. As would be

expected, the differences between the estimates based on the non-institutionalized and estimates

based on the entire population are largest for the least educated and smallest for the most

educated. In 2000, including the institutionalized increases the white-black employment rate

differential by six percentage points for high school dropouts, four points for high school

graduates, three points, for those with some college education, and one points for those with at

least a college degree.

The effect of increased incarceration on estimates of black-white wage convergence is

likely to impart less of a selection bias than that created by the more general decline in black

8 Katz and Krueger (1999) estimate that the decline in the national unemployment rate of 2.6 percentage points between 1985 and 1998 would have been 0.1 to 0.5 points lower had incarceration rates not increased over this time period.

14

employment rates. That is to say, such a large fraction of non-institutionalized black men are

non-employed in any given week that incarceration is unlikely to select out a substantial fraction

of wage earners. Thus, researchers interested in black-white wage trends have focused more

generally on estimating the selection effect of declining relative employment rates on aggregate

black-white wage ratios (accounting for both the incarcerated and the non-incarcerated non-

employed).

To the extent that black labor market dropouts are disproportionately concentrated in the

lower tail of the wage-offer distribution, the declining relative employment rate of black men

will bias aggregate wage trends towards convergence. Assessing this selection bias requires

recovering the wage-offer distribution for labor market dropouts and recalculating measures of

the central tendency of wage offers taking into account the entire distribution (both observed

offers for the employed and the unobserved offer distribution for the non-employed) (Chandra

2003). Several authors have attempted to tackle this problem. Brown (1984) provides one of the

earliest examples. Brown assumes that the non-institutionalized, non-employed of each racial

group would earn wages below their group-specific median, and calculates hypothetical trends in

black-white median wage ratios based on this assumption.10 The author argues that the majority

of observed wage convergence during the 1970s is directly attributable to the declines in black

labor force participation rates.

9 Here, I classify as employed those who have a job and those who are in the armed forced. Excluding the armed forces from these calculations has little impact on the estimates in Figures 3 through 7. These calculations make use of the 1970, 1980, 1990, and 2000 one percent PUMS data. 10 This strategy is also employed by Neal and Johnson (1996) in their analysis of the effect of adjusting for scores on the Armed Forced Qualifying Test (AFQT) on estimates of residual racial wage differentials. Ignoring labor market dropouts, Neal and Johnson find that accounting for AFQT scores explains nearly all of the residual wage differential between black and white workers observed in the National Longitudinal Survey of Youth. However, in least-absolute-deviation regressions that assign wage values of zero to all non-participants, a substantial AFQT-adjusted racial wage differential reappears.

15

More recent efforts to account for both the non-institutionalized jobless as well as the

incarcerated are offered by Juhn (2003) and Chandra (2003). Juhn assigns the average wage of

similarly situated employed workers (of similar race, age, and education) to the non-employed

and recalculates relative black-white wage trends with and without this adjustment. Juhn finds

that while wage ratio estimates that are not adjusted for selection bias find convergence during

the 1960s and 1970s, and stagnation thereafter, adjusting for selection bias reveals a widening of

racial wage differentials during recent decades.

The most thorough research on this question is provided by Chandra (2003). Chandra

presents a series of alternative adjustments for selection bias, including:

• a matching estimator that assigns the mean earning for comparable workers to the non-

employed (a la Juhn),

• a matching estimator that assigns the median earnings for comparable workers to the non-

employed,

• calculating race-age-education cell-specific median earnings base on the assumption that

non-participants are negatively selected from the offer distribution within cells and then

calculating overall medians based on these tabulations, and

• modifying the selection specification in the previous bullet so that only the long-term non-

employed are negatively selected within group.

All four selection-corrections find that black-white wages diverged during the 1980s, with the

methods based on the assumptions that labor market dropouts are negatively selected within

16

group yielding the largest increases in racial wage differentials. Moreover, Chandra shows that

incarceration contributed significantly to the divergence during the 1980s.11

3. The Effect of Incarceration on Future Labor Market Prospects

Serving prison time is likely to adversely affect one’s labor market prospects for a number of

reasons. First, a prison term interrupts one’s work career. Incarcerated felons cannot accumulate

employment experience while serving time. In addition, the quality of one’s non-

institutionalized social network, a main source of employment information, is likely to erode

with time incarcerated.

Second, serving time in prison is stigmatizing. Employers consistently express reservations

about applicants with prior criminal history records. Many employers that offer low-skilled jobs

either ask about criminal history records or perform formal reviews of applicant criminal history

records. In some instances, employers are prohibited from hiring convicted felons.

Given the trends documented above, these adverse effects of incarceration are sure to be an

increasingly important determinant of black employment outcomes. This section explores these

factors in detail. I begin by documenting the extent to which a prison term interrupts the

potential legitimate work careers of young men. Using administrative data for California, I chart

the incarceration experience during the 1990s of young felons that began serving time on a

prison commitment at any point during the year 1990. Next, I review what is known about

employer sentiments with regards to ex-offenders and the extent to which employers formally

and informally screen-out felons from the applicant pool.

To what extent does prison interrupt one’s potential work career?

11 The study does not provide an analysis of wage trends during the 1990s. Additional attempts to assess the importance of selection bias to estimates of trends in black-white wage ratios is provided by Heckman, Lyons, and

17

The extent to which being sentenced to prison interrupts a felon’s potential work career

depends on both the expected amount of time served on a typical term as well as the likelihood

of serving subsequent terms. Newly admitted prisoners during the late 1990s were generally

serving time on commitments with a maximum sentence of three years and a minimum sentence

of one year (with many serving time closer to the minimum) (Raphael and Stoll 2003). If this

were the only time served for most felons, than the time interruption of prison would not be that

substantial for most.12

However, many felons serve multiple terms in prison, either due to the commission of

new felonies or due to violation of parole conditions after being released from the initial spell. A

large body of criminological research consistently finds that nearly two thirds of ex-inmates are

rearrested within a few years of release from prison (Petersilia 2003). Moreover, a sizable

majority of the re-arrested will serve subsequent prison terms. Thus, for many offenders, the

period of time between the ages of 18 and 30 is characterized by multiple short prison spells with

intermittent, and relatively short, spells outside of prison.

Here, I document the prison histories of young offenders entering the California state

prison system. While there are several reasons to suspect that California may not be

representative of the nation,13 there are also several reasons why such an exercise is instructive.

First, the state incarceration rate in California (453 per 100,000) is near the national average (422

per 100,000), and thus at least along this dimension, the state is comparable to the nation.

Second, the California state prison system is the largest in the nation (accounting for 13 percent

of the nation’s state prison population), and thus the experience of California inmates is an

Todd (2000) and Smith and Welch (1986). 12 Of course, we are not saying that a year in prison is not costly. However, a year absence from the labor market during the beginning of one’s career would have only a small effect on accumulated experience.

18

important contributor to the weighted-average experience. Finally, these detailed administrative

records permit linking subsequent spells and characterization of personal prison histories,

something that can’t be done with publicly available inmate surveys.

I focus on offenders between 18 and 25 that enter one of the state’s prisons on a new

court commitment (that is to say, not for a parole violation) during the year 1990, and

characterize several aspects of their prison experiences over the subsequent decade. The analysis

is based on administrative records obtained from the California Department of Corrections for all

prison terms served in the state with a start date during the 1990s. Figures 4 and 5 present the

relative frequency distributions for the number of separate prison terms served by our cohort,

inclusive of the initial term caused by the 1990s court commitment and any subsequent terms

due to either additional court commitments or parole revocations. Figure 4 presents the

distribution for all inmates while Figure 5 shows separate distributions by race and ethnicity.14

The majority of inmates serve more than one prison term. Only 32 percent of inmates entering

prison during the year 1990 serve one term during the subsequent decade. The median inmate

serves two terms over the decade, while a substantial fraction (approximately 48 percent) serves

three or more terms.

There are notable differences by race and ethnicity. Only 28 percent of white inmates

and 23 percent of black inmates serve one term during the decade, while the comparable figure

for Hispanic inmates is nearly 40 percent. The median white and black inmates serve three terms

while the median Hispanic inmate serves two terms. Again, substantial fractions of this 1990

13 In particular, California’s high rate of parole violation and the fact that nearly all released inmates are released to parole status is unique among the 50 states. 14 White, Hispanic, and Black inmates each account for approximately one third of the California state prison population. Asians and members of other racial groups account for a small proportion of inmates.

19

cohort serve three or more terms, with 54 percent of white inmates, 58 percent of black inmates,

and 39 percent of Hispanic inmates entering prison at least three times.

In California, the time served on each term tends to be relatively short. While there are

many California inmates serving long sentences and several thousand that are serving 25 years to

life under the state’s “three strikes” sentencing provision, the majority of inmates are sentenced

to relatively short terms that are often completed within a year or two. Table 6 shows the median

time served for inmates in our 1990 cohort. The table provides figures for all inmates, by

race/ethnicity, and by whether the term is the first, second, third, fourth, or fifth or higher term.

For the first term served (which, for our sample, also corresponds to the first term served on a

specific court commitment), the median time served for all inmates is approximately one year.

The median time for white inmates is somewhat lower (0.94 years), the median for black inmates

is somewhat higher (1.17 years), while the median for Hispanics is equal to the overall median.

Median time served declines uniformly with subsequent terms served, declining to 0.68

years for the second term, 0.62 years for third, and so one. This decline reflects the fact that

many of these subsequent prison terms are served for parole violations rather than new felony

court commitments, and thus represent time served on the remaining sentence from the initial

court commitment sending the offender to prison in the first place.

The figures in Table 6 suggest that for most felons committed to prison the actual amount

of time behind bars is fairly short, even accounting for the likelihood of serving multiple spells.

Panel A of Table 7 presents estimates of the total amount of time served accounting for multiple

terms (summing time served across all terms for each inmate) for the inmate at the 25th, 50th and

75th percentiles of this distribution. The median inmate spends 2.8 years during the 1990s in one

of California’s state prisons, with the median white inmate (3.09 years) and median black inmate

20

(3.53 years) serving more time and the median Hispanic inmate (2.23 years) serving less time.

Roughly 25 percent of inmates served at least 5 years during the 1990s while another 25 percent

served less than 1.5 years.

However, as a gauge of the extent to which incarceration potentially interrupts the

accumulation of legitimate labor market experience, these figures are misleading. Most

importantly, cumulative time served does not account for the short periods of time between

prison spells where inmates may find employment, yet are not able to solidify the employment

match with any measurable amount of job tenure. A more appropriate measure of the degree to

which incarceration impedes experience accumulation would be the time between the date of

admission to prison for the first term served and the date of release from the last term.

Panel B of Table 7 presents the quartile values from the distribution of this variable. For

the median inmate, five years elapses between the first date of admission and the last date of

release. For median white, black, and Hispanic inmates, the comparable figures are 6.2, 6.5, and

3.2 years, respectively. For approximately one quarter of inmates (and more than one quarter of

white and black inmates), nine years pass between their initial commission to prison and their

last release. In other words, one quarter of these inmates spends almost the entire decade cycling

in and out of prison.

Moreover, the estimates of the work life interruption in Panel B of Table 7 are likely to

be lower bound estimates. Our cohort is comprised of 18 to 25 years old inmates that enter the

state prison sentence on a new court commitment in 1990. Surely, many of the older offenders

in this group have served prior prisons sentences on previous commitments, a factor that would

add to our estimates if we had information on earlier time served. Moreover, many of the

younger offenders are likely to have juvenile records and may have been previously incarcerated

21

in the California Youth Authority system. In addition, we do not have data on time served for

commitments with start dates occurring after 1999 –i.e., our panel arbitrarily places a cap on the

amount of time one can be involved with the prison system. Finally, we cannot observe time

served in jail while awaiting trial or a parole revocation hearing. Accounting for the jail time

that usually accompanies the transition between being non-institutionalized and imprisonment

would surely increase these estimates.

Regardless, spending 5 years of one’s early life (6.5 years for the median black offender)

cycling in and out of prison must impact one’s earnings prospects. Clearly, being behind bars

and the short spans of time outside of prison prohibit the accumulation of job experiences during

a period of one’s life when the returns to experience are the greatest.

Does having been in prison stigmatize ex-offenders?

The potential impact of serving time on future labor market prospects extends beyond the

failure to accumulate work experience. There is considerable evidence that employers are averse

to hiring former prison inmates and often use formal and informal screening tools to weed ex-

offenders out of the applicant pool. Given the high proportion of low-skilled black men with

prison time on their criminal history records, such employer sentiments and screening practices

represent an increasingly important employment barrier for African-American males. Moreover,

as well we discuss below, this stigmatization of prisoners, coupled with the informal screening

methods used by employers, may also be adversely impacting the employment prospects of

young black men without criminal history records.

Employers consider criminal history records when screening job applicants for a number

of reasons. For starters, certain occupations are closed to felons under state, and in some

instances, federal law (Hahn 1991). Examples include jobs requiring contact with children,

22

certain health services occupations, public employment in some states and localities, and

employment in firms providing security services. In addition, in many states employers can be

held liable for the criminal actions of their employees. Under the theory of negligent hiring,

employers can be required to pay punitive damages as well as damages for loss, pain, and

suffering for acts committed by an employee on the job (Craig 1987). Finally, employers

looking to fill jobs where employee monitoring is imperfect may place a premium on

trustworthiness. To the extent that past criminal activity signals a lack of trustworthiness,

employers may take such information into account when screening applicants.

In all known employer surveys where employers are asked about their willingness to hire

ex-offenders, employer responses reveal a strong aversion to hiring applicants with criminal

history records (Holzer, Raphael, and Stoll 2002, 2003; Pager 2003). For example, Figure 6

present tabulations from the employer survey of the Multi-City Study of Urban Inequality

(MCSUI).15 The figure presents employer responses to questions inquiring about the likelihood

that the employer would hire various types of job applicants, including applicants with a criminal

history record. Over 60 percent indicated that they would “probably not” or “definitely not” hire

applicants with criminal history records, with “probably not” being the modal response. Since

these data pertain to employers that have recently hired low-skilled workers (employer who are

perhaps the most likely to employ an ex-offender), these results imply that a large majority of

employers are unwilling to hire former prison inmates.

Employer aversion to applicants with criminal history records is stronger than employer

aversion to hiring other groups of commonly stigmatized workers. Relative to those with

criminal history records, employers are considerably less averse to hiring the other types of

23

applicants displayed in Figure 6. For example, while 60 percent of employers indicate that they

are unlikely to hire an ex-offenders,16 the comparable figures are eight percent for welfare

recipients, three percent for applicants with a GED, 41 percent for applicants with spotty work

histories, and 18 percent for applicants unemployed for over a year. Interestingly, employers

exhibit the stronger reluctance to hiring applicants with spotty work histories, a characteristic

that one might interpret as signaling past incarceration. However, even for this group of

applicants, the proportion of employer unwilling to hire such workers is 75 percent of the

proportion unwilling to hire ex-offenders.17

The ability of employers to act on an aversion to ex-offenders, and the nature of the

action in terms of hiring and screening behavior, will depend on employer accessibility to

criminal history record information. To the extent that an employer can and does access criminal

history records, an employer may simply screen out applicants based on their actual arrest and

conviction records. Among the employers interviewed in the MCSUI sample, 32 percent

indicated that they always check the criminal history records of applicants, 17 percent indicate

that they sometime check, while 51 percent indicate that they never check. More recent

employer surveys (the MCSUI data are from the early 1990s) indicate that the use of formal

background check has increased. For example, a 2001 survey of employers in Los Angeles with

questions similar to those in the earlier MCSUI survey found that 46 percent of employers

indicates that they always perform criminal background checks, 18 percent indicated that they

15 These data where collected in 1993 and 1994 from establishments in the Atlanta, Boston, Detroit, and Los Angeles metropolitan statistics areas. See Holzer, Raphael, and Stoll (2002) for a complete description of this survey. 16 We define unlikely to hire as either responding that one would “probably not” or “definitely not” be willing to hire an applicant with the given characteristic. 17 Using a more detailed survey of employer preferences, Holzer, Raphael, and Stoll (2003) uncover a fair degree of nuance in employer sentiments with regards to ex-offenders. For example, employers are willing to consider such factors as specific offense and the amount of time lapsed since the offense committed in evaluating applicants with criminal history records.

24

sometimes checked, while 37 percent indicated that they never checked.18 The comparable

tabulations for Los Angeles from the earlier MCSUI data indicated that 32 percent always check,

16 percent sometimes check, while 52 percent never check.

In the absence of a formal background check, an employer may act on their aversion to

hiring ex-offenders using perceived correlates of previous incarceration, such as age, race, or

level of educational attainment to screen out those with criminal histories. In other words,

employers may engage in statistical discrimination against applicants that are thought to come

from demographic groups with high rates of involvement in the criminal justice system. Holzer,

Raphael, and Stoll (2002) find employer-hiring patterns consistent with such statistical

discrimination against black men. Specifically, the authors find that employers that check

criminal history records are more likely to hire blacks than employers that do not, and that this

positive effect of criminal background checks is strongest among those employers with the

strongest stated aversion to hiring ex-offenders.19 Via such statistical discrimination, even young

black men who have never been incarcerated or involved with the criminal justice system are

harmed by the rise in incarceration among young black men overall.

The study by Pager (2003) offers perhaps the clearest evidence of employer aversion to

ex-offenders and the stigma associated with having served time in prison. Pager conducted an

audit study of employer hiring activity in the year 2001 for employers located in the Milwaukee

metropolitan area. Using male auditors matched on observable characteristics including age,

education, general appearance, demeanor, and race, Pager assesses the effect of prior prison

experience on the likelihood that each auditor is called back for an interview. The auditor

18 See Holzer, Raphael, and Stoll (2003) for a description of this latter survey and the results. 19 In addition, the authors find that this relative effect for the hiring of black men, but not black women, and for employer willingness to hire applicants from other stigmatized groups, especially workers with gaps in their employment histories.

25

assigned the criminal history record signaled having served time by indicating six months of

prison work experience on his resume.

Figure 7 presents the main results from this study. For both black and white auditors, the

call back rate for the auditor with a criminal history record is less than half the call back rate for

the auditor without the criminal history record. Interestingly, the call back rate for blacks that do

not signal prior prison time is less than the call back rate for whites with a prior prison term.

Such a pattern is consistent with either taste-based discrimination or employers statistically

discriminating against black applicants without a signaled record in an attempt to avoid ex-

offenders. Regardless, the results presented by Pager are stark and indicative of the employment

barriers faced by ex-offenders.

4. Increasing Incarceration Rates and the Decline in Black Employment Rates

So far, we have documented several stylized facts. First, the rate of current incarceration

among black men has increased considerably over the past three decades, with particularly

pronounced increases among prime-age, less-educated black men. Second, the proportion of

black men with prison time on their criminal history records has increased by even greater

amounts. Third, serving time in prison substantially interrupts the potential work lives of young

offenders, leading to at least six years of lost labor market experience for the median black

offender in California, and more time for a sizable minority of offenders. Fourth, employers are

averse to hiring ex-inmates, and use formal and informal screening methods to weed such

individuals out of the applicant pool.

In addition to these criminal justice trends, we have also demonstrated that the

employment rate of non-institutionalized black men has declined over the same period. Between

26

1970 and 2000 the white-black employment rate differential widened from eight percentage

points to 17 percentage points among non-institutionalized men. For high school dropouts, this

differential increased from seven to 19 percentage points.

The preceding discussion suggests at least two avenues by which the increase in black

incarceration rates may be related to the relative decline in black employment rates. First, with

an increasing incarceration rate, the proportion of non-institutionalized black men with prior

prison time has increased. These men have less legitimate labor market experience than

otherwise similar men who have not been to prison and must contend with strong employer

reluctance to hire ex-offenders. Second, men without criminal histories who are members of

demographic groups where a large proportion has a criminal history record may face statistical

discrimination in the labor market and difficulty finding employment.

In this section, I test for a partial correlation between the proportion of a given sub-group

of men that are institutionalized and the proportion of comparable non-institutionalized men that

are employed. These estimates are then used to provide an estimate of the proportion of the

black-white employment rate differential among the non-institutionalized that may be

attributable to incarceration trends.

Using data from the 1970, 1980, 1990, and 2000 PUMS, I first estimate the proportion of

non-institutionalized men that are employed and the proportion of all men that are

institutionalized for 320 demographic sub-groups. The sub-groups are defined by interaction of

four mutually exclusive race/ethnic groups, five age groups, four education groups, and four

years. I use the five age groups and four education groups listed in Tables 1 and 2, and compute

separate figures for non-Hispanic whites, non-Hispanic blacks, non-Hispanic Asians, and

Hispanics. I then estimate a series of regression models where the key dependent variable is the

27

proportion of non-institutionalized men that are employed and the key explanatory variable is the

proportion of all men from the given demographic group that are institutionalized.

Table 8 presents these estimation results. Each model regresses the non-institutionalized

employment rate on dummy variables indicating the race/ethnicity of the group and interaction

terms between these dummies and a set of year dummies. The coefficients on these dummies

provide estimates of the employment rate differentials relative to whites in 1970 and how these

differentials change across decades. The first three regression models include a complete set of

dummy variables indicating the age/education cell. The last three regressions include a complete

set of dummies indicating the age/education/year cells (thus allowing the effect on employment

of being in a specific age/education group to change over time). Within these two groups of

regressions, the first specification omits the variable measuring the proportion institutionalized,

the second specification adds the institutionalization variables, while the final specifications adds

the institutionalization variable with a full set of interactions with the year dummies.20

Within both sets of regressions (those including and those omitting the

age/education/year interaction terms), adding the proportion institutionalized reduces the

coefficients on the interaction terms between black and the year dummies. In other words, the

proportion institutionalized explains part of the widening of the black-white employment rate

differential in all decades between 1970 and 2000. For example, the results in regression model

(1) indicate that after adjusting for age/education effects and year effects, the black-white

employment rate differential widened by 12 percentage points between 1970 and 2000. In

contrast, the comparable estimate from regression (2) indicates a widening of 6.4 percentage

20 Each regression is weighted by cell frequency. In addition, each regression includes a constant term and a set of base year effects.

28

points. The findings are similar in the models where the age/education dummies are interacted

with year.

The proportion institutionalized has a strong negative effect on the proportion of non-

institutionalized men that are employed. In the models where the institutionalization variable is

interacted with year, the proportion institutionalized has the largest effect on employment rates in

1980, followed by 1990, and 2000. Again, the regression results are comparable when the

age/education/year interaction terms are included.

Figure 8 summarizes the effect of controlling for the proportion institutionalized on the

white-black difference in employment rates among non-institutionalized men. The employment

rate differentials omitting controls for institutionalization come from the model (4) regression

results while the differentials adjusting for institutionalization comes from the model (6)

regression results. Adjusting for institutionalization reduces the residual employment rate

differential in all years, although the largest reductions occur in 1990 and 2000. In 1990,

adjusting for institutionalization reduces the residual difference from 12.7 percentage points to

5.3 percentage points. In 2000, adjusting for institutionalization reduces this differential from

15.5 percentage points to 10.1 percentage points.

Thus, this very simple exercise yields a quite high upper bound contribution of changes

in incarceration rates to recent employment trends. To be sure, there are problems with this

approach to the question. In particular, the proportion institutionalized within a given group is

likely to be correlated with unobservable characteristics of men that vary within age/education

cells and that determine their employability. Moreover, while the variation used to identify the

effect of the institutionalization variable accounts for cross-year changes in the employment rates

of the age/education sub-groups, these employment rate differentials may have changed within

29

racial and ethnic groups, and in manner correlated with changes in incarceration rates.21

Nonetheless, the results suggest that incarceration trends may explain a large portion of currently

observed racial employment rate differentials.

5. Conclusion

This paper has documented several trends in the incarceration of black men and highlighted

the potential lasting effects of high incarceration rates on their future employment prospects and

on their relative socioeconomic status. To summarize, the paper has demonstrated that

• the current incarceration rates of black men are extraordinarily high by historical comparison

especially for less-educated and relatively young black men,

• the proportion of black men ever having served time is even higher,

• prison time substantially interrupts the potential work careers in the legitimate labor market

of young felons, and

• employers are extremely reluctant to hire applicants who have served time in state or federal

prison.

In addition, I have presented simple estimates suggesting that a sizable portion of the black-white

employment rate differentials may be attributable to the high rate of involvement of blacks with

the criminal justice system. Across demographic groups defined by education, age, and race,

there is a strong inverse correlation between the proportion of the group currently

institutionalized and the employment rate of the non-institutionalized. Moreover, this partial

21 An alternative estimation strategy that may be used to address this issue would be to calculate the 320 sets of means for each state, and then include a full set of race/age/education/year fixed effects in the final specification. I experimented with this approach and found that the sample sizes using the one percent PUMS yielded extremely small cells for many states and thus, quite imprecise estimates. Revisiting this exercise with the 5 percent samples, however, may solve this problem.

30

correlation is strong enough to explain roughly one-third of the black-white employment rate

differential in 2000.

The continual increase in the lifetime likelihood of going to prison between 1970 and

2000 suggests that the proportion of blacks with prior prison time is likely to increase in the

coming years, even if current incarceration rates remain unchanged. This follows from the fact

that the risk of incarceration is highest early in one’s adult life, and that young men coming of

age today face higher incarceration rates than those faced by comparable young men during the

1980s and 1990s. Thus, the barriers faced by ex-offenders are likely to hamper the

socioeconomic progress of many black men for the foreseeable future.

Clearly, this is a topic that deserves future attention from both researchers and policy

makers. There are several potentially fruitful directions that researchers can pursue. While the

incarceration trends documented here are often attributable to changes in sentencing policy and

potential changes in offending among young men, there are few systematic evaluations of the

relative culpability of behavioral trends and key policy choices. For example, there is little

evidence on the relative impact of state vs. federal sentencing reforms on black incarceration

rates. At the state level, the shift from indeterminate to determinate sentencing, sentence

enhancements for drug-related crime such as those enacted under Governor Nelson Rockefeller

in New York, three-strikes laws, and state-level sentence enhancement for violent crimes and

crime committed with a firearm are all likely contributors. At the federal level, differential

sentences for crack vs. powder cocaine, systematic efforts to try felons in possession of firearm

in federal courts, and the federal sentencing reforms that limited judicial discretion are an

additional set of potential candidates. Assessing the extent to which these state and federal

sentencing changes are having a disparate impact on blacks would be most useful. Such research

31

may be of help in deciding how to deploy criminal justice resources more efficiently and in

assessing whether current sentencing policy is unnecessarily impacting the long-term

employment prospects of an already disadvantaged group.

In addition, while there are many qualitative evaluations of the effects of prisoner education

and in-prison treatment programs, careful empirical studies employing rigorous experimental or

quasi-experimental research designs are few and far between. Early non-experimental

evaluations of state employment programs tailored at paroled ex-offenders (reviewed in

Bushway and Reuter (2002)) arrived at quite pessimistic conclusions regarding the ability of

training and job search assistance to lower the recidivism of parolees. Evaluations of more

recent state programs, however, are uniformly more positive. In a review of recent research,

Petersilia (2002) cites several evaluations finding program effects on the likelihood that parolees

find employment on the order of 20 percentage points, and effects on the likelihood of re-arrest

and being returned to prison custody on the order of 10 percentage points. A review of 19 studies

by Wilson et. al. (2000) finds similar program effects.22

Careful reviews of the more recent evaluations tend to attribute the high estimates from the

latter research to flaws in methodological design. Bushway and Reuter (2002) as well as Wilson

et. al. (2000) note that few of the program evaluations are based on randomized designs where

program participation is determined by random assignment rather than self-selection. Moreover,

many of the evaluation do not control for differences in offender characteristics that may

simultaneously explain program participation and recidivism/parole violation rates. Thus, there

is much room for additional work.

22 For detailed descriptions of three state level job training and placement programs, see Finn (1998a, 1998b, and 1998c).

32

In general, prisoner reentry policy is and will be an important determinant of the relative

socioeconomic status of low-skilled black men. There is a great need for creative thinking on

how to combine existing services for the economically disadvantaged (such as workforce

development programs) with services targeted towards the needs of ex-offender in a manner that

will minimize the negative collateral consequences of prior prison time.

33

References

Beck, Allen J.; Karberg, Jennifer C. and Paige M. Harrison (2002), Prison and Jail Inmates at Midyear 2001, Bureau of Justice Statistics Bulletin, NCJ 191702. Bonczar, Thomas P. (2003), Prevalence of Imprisonment in the U.S. Population, 1974-2001, Bureau of Justice Statistics Special Report, NCJ 197976. Bound, John and Richard B. Freeman (1992), “What Went Wrong? The Erosion of the Relative Earnings and Employment of Young Black Men in the 1980s,” Quarterly Journal of Economics, 107(1): 201-232. Brown, Charles C. (1984), “Black-White Earnings Ratios Since the Civil Rights Act of 1964: The Importance of Labor Market Dropouts,” Quarterly Journal of Economics, 95(1): 31-44. Bushway, Shawn and Peter Reuter (2002), “Labor Markets and Crime,” in Wilson, James Q. and Joan Petersilia (eds.) Crime: Public Policies for Crime Control, Oakland: CA, Institute for Contemporary Studies Press, 191-224. Butcher, Kristin F. and Anne Morrison Piehl (1998), “Recent Immigrants: Unexpected Implications for Crime and Incarceration,” Industrial & Labor Relations Review, 51(4): 654-679. Chandra, Amitabh (2003), “Is the Convergence in the Racial Wage Gap Illusory?” National Bureau of Economic Research Working Paper #9476. Craig, Scott R. (1987), “Negligent Hiring: Guilt By Association,” Personnel Administrator, October: 32-34. Finn, Peter (1998a), Chicago’s Safer Foundation: A Road Back for Ex-Offenders, Washington, D.C., National Institute of Justice. Finn, Peter (1998b), Successful Job Placement for Ex-Offenders: The Center for Employment Opportunities, Washington, D.C., National Institute of Justice. Finn, Peter (1998c), Texas’ Project RIO (Re-integrating of Offenders), Washington, D.C., National Institute of Justice. Hahn, J.M. (1991), “Pre-Employment Information Services: Employers Beware,” Employee Relations Law Journal, 17(1): 45-=69. Heckman, James J.; Lyons, Rom and Petra Todd (2000), “Understanding Black-White Wage Differentials,” American Economic Review, 90(2): 344-349. Holzer, Harry J. and Paul Offner (2003), “Trends in the Employment Outcomes of Young Black Men, 1979 –2000,” in Ronald Mincy (ed), Young Black Men.

34

Holzer, Harry J.; Raphael, Steven and Michael A. Stoll (2002), “Perceived Criminality, Racial Background Checks, and the Racial Hiring Practices of Employers,” working paper. Holzer, Harry J.; Raphael, Steven and Michael A. Stoll (2003), “Employer Demand for Ex-Offenders: Recent Evidence from Los Angeles,” Working Paper. Juhn, Chinhui (2003), “Labor Market Dropouts and Trends in the Wages of Black and White Men,” Industrial & Labor Relations, 56(4): 643-662. Katz, Lawrence F. and Alan B. Krueger (1999), “The High-Pressure U.S. Labor Market of the 1990s,” Brookings Papers on Economic Activity 1999, 1-87. Neal, Derek and William Johnson (1996), “The Role of Pre-Market Factors in Black-White Wage Differences, Journal of Political Economy, 104(5): 869-895. Pager, Devah (2003), “The Mark of a Criminal Record,” American Journal of Sociology, 108(5): 937-975. Petersilia, Joan (2002), “Community Corrections,” in Wilson, James Q. and Joan Petersilia (eds.) Crime: Public Policies for Crime Control, Oakland: CA, Institute for Contemporary Studies Press, 483-508. Petersilia, Joan (2003), When Prisoners Come Home: Parole and Prisoner Reentry, Oxford University Press: Oxford. Raphael, Steven and Michael Stoll (2004), “The Effect of Prison Releases on Regional Crime Rates,” forthcoming in William G. Gale and Janet Rothenberg Pack (eds), The Brookings-Wharton Papers on Urban Economic Affairs, Volume 5, The Brookings Institution: Washington. Raphael, Steven and David Weiman (2003), “The Impact of Local Labor Market Conditions on the Likelihood That Parolees are Returned to Custody,” Working Paper. Smith, James P. and Finis Welch (1986), Closing the Gap: Forty Years of Economic Progress for Blacks, Rand Corporation: Santa Monica, CA. U.S. Census Bureau (2003), Technical Assessment of A.C.E. Revision II, Accessed on March 17, 2004, http://www.census.gov/dmd/www/ace2.html. Western, Bruce and Becky Petit (2000), “Incarceration and Racial Inequality in Men’s Employment,” Industrial & Labor Relations Review, 54(1): 3-16. Wilson, David B.; Gallagher, Catherine. A.; and Doris L. MacKenzie, (2000), “A Meta-Analysis of Corrections-Based Education, Vocation, and Work Programs for Adult Offenders, Journal on Research in Crime and Delinquency, 37: 347-368.

35

Figure 1: Comparison of BJS Estimates of the Number of Men in Federal Prison, State Prison, and Local Jails (2001) to Estimates from the 2000 One Percent Census Public Use Microdata Sample of the Number of Men Institutionalized by Race and Ethnicity

6 8 4 ,8 0 0

8 0 3 ,4 0 0

2 8 3 ,0 0 0

7 8 1 , 9 2 38 1 3 ,9 2 0

3 2 2 ,9 6 0

0

1 0 0 ,0 0 0

2 0 0 ,0 0 0

3 0 0 ,0 0 0

4 0 0 ,0 0 0

5 0 0 ,0 0 0

6 0 0 ,0 0 0

7 0 0 ,0 0 0

8 0 0 ,0 0 0

9 0 0 ,0 0 0

N o n -H is p a n ic W h it e N o n -H is p a n ic B la c k H is p a n ic

Num

ber o

f Men

B u re a u o f Ju s t ic e S ta t is t ic s2 0 0 0 1 % P u m s

Figure 2: Lifetime Likelihood of Serving a State or Federal Prison Sentence for Males Born in 1974 and 2001 by Race and Ethnicity

2.2%

13.4%

4.0%

5.9%

32.2%

17.2%

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

35.0%

Non-Hispanic W hite Non-Hispanic B lack Hispanic

Life

time

Like

lihoo

d of

Goi

ng to

Pris

on

19742001

36

Figure 3: White-Black Employment Rate Differentials for the Non-Institutionalized and the Non-Institutionalized and Institutionalized Combined

0.08

0.130.14

0.17

0.09

0.15

0.17

0.22

0.00

0.05

0.10

0.15

0.20

0.25

0.30

1970 1980 1990 2000

Empl

oym

ent r

ate

diffe

renc

e, w

hite

-bla

ck

Non-Ins t itu t ionaliz edA ll

Figure 4: The Distribution of 18 to 25 Year Old Prisoners Entering the California State Prison System in 1990 by the Number of Terms Served over the Subsequent Decade

31.99%

19.79%

14.06%

10.22%

7.52%

5.42%

4.00%2.65%

4.34%

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

1 2 3 4 5 6 7 8 9+

Number of Terms Served

Perc

ent o

f 199

0 Pr

ison

er C

ohor

t

37

Figure 5: The Distribution of 18 to 25 Year Old Prisoners Entering the California State Prison System in 1990 by the Number of Terms Served over the Subsequent Decade, by Race and Ethnicity

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

40.00%

45.00%

Num be r of Te rm s S e rve d

Perc

ent o

f 199

0 Pr

ison

er C

ohor

t

W hitesB lacksHispanic

W hites 27.63% 16.26% 12.37% 11.51% 9.09% 7.61% 5.44% 3.58% 6.51%

B lacks 23.95% 18.90% 16.13% 11.61% 9.18% 6.45% 4.77% 3.67% 5.35%

Hispanic 39.25% 22.16% 13.41% 8.95% 5.55% 3.64% 2.88% 1.58% 2.57%

1 2 3 4 5 6 7 8 9+

Figure 6: Self-Reported Employer Willingness to Hire Applicants from Various Groups from the Establishment Survey of the Multi-City Study of Urban Inequality

13%

52%

57%

19%

29%

26%

40%39% 40%

54%

42%

7%

2%

35%

16%

20%

1% 1%

6%

2%

0%

10%

20%

30%

40%

50%

60%

70%

Crim inal Record W elfare Rec ipient GED no High SchoolDiploma

Spotty W ork His tory Unemployed for Morethan a Year

Perc

ent o

f Em

ploy

ers

Definitely W illP robably W illP robably NotDefinitely Not

38

Figure 7: Percent of Auditors Called Back for an Interview by Race and Whether the Auditor Was Assigned a Criminal History Record, as Reported in Pager (2003)

14%

34%

5%

17%

0%

5%

10%

15%

20%

25%

30%

35%

40%

Black W hite

Perc

ent R

ecei

ving

a C

all B

ack

W ithout RecordW ith Crim inal His tory Record

Figure 8: White-Black Difference in the Proportion Employed Among the Non-Institutionalized, With and Without Controlling for the Proportion of the Population Institutionalized

0.058

0.118

0.127

0.155

0.045

0.064

0.053

0.101

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

1970 1980 1990 2000

Empl

oym

ent R

ate

Diff

eren

tial

W ithout Controlling for the ProportionInstitutionalizedControlling for the Proportion Instituionalized

Table 1 Employment and Institutionalization Status for Non-Hispanic Black Males and Non-Hispanic White Males by Educational Attainment, 1970 to 2000 Black Males White Males Age 1970 1980 1990 2000 1970 1980 1990 2000All Employed NILF Armed Forces Institutionalized

0.73 0.20 0.04 0.03

0.64 0.29 0.04 0.03

0.63 0.30 0.03 0.04

0.57 0.33 0.02 0.08

0.82 0.13 0.04 0.01

0.80 0.17 0.02 0.01

0.80 0.17 0.02 0.01

0.79 0.18 0.01 0.01

Less than high school Employed NILF Armed Forced Institutionalized

0.71 0.23 0.01 0.04

0.57 0.38 0.01 0.05

0.46 0.44 0.00 0.10

0.34 0.47 0.00 0.19

0.80 0.17 0.01 0.02

0.69 0.28 0.01 0.02

0.63 0.34 0.00 0.03

0.59 0.37 0.00 0.04

High school graduate Employed NILF Armed Forced Institutionalized

0.75 0.15 0.08 0.02

0.66 0.25 0.06 0.03

0.63 0.28 0.04 0.05

0.56 0.35 0.02 0.08

0.85 0.10 0.05 0.01

0.81 0.15 0.03 0.01

0.80 0.17 0.02 0.01

0.77 0.20 0.01 0.02

Some College Employed NILF Armed Forced Institutionalized

0.71 0.22 0.06 0.01

0.69 0.25 0.04 0.02

0.68 0.23 0.05 0.05

0.66 0.26 0.03 0.05

0.77 0.18 0.04 0.00

0.81 0.17 0.02 0.00

0.81 0.16 0.02 0.01

0.80 0.17 0.02 0.01

College Plus Employed NILF Armed Forced Institutionalized

0.87 0.09 0.04 0.01

0.84 0.13 0.02 0.01

0.85 0.11 0.03 0.01

0.81 0.16 0.01 0.01

0.89 0.07 0.04 0.00

0.91 0.07 0.02 0.00

0.90 0.08 0.02 0.000

0.89 0.10 0.01 0.00

Figures are tabulated from the 1970, 1980, 1990, and 2000 public use micro data samples from the U.S. Census of Population and Housing.

Table 2 Employment and Institutionalization Status for Non-Hispanic Black Males and Non-Hispanic White Males by Age, 1970 to 2000 Black Males White Males Age 1970 1980 1990 2000 1970 1980 1990 200018 to 25 years Employed NILF Armed Forced Institutionalized

0.55 0.32 0.08 0.05

0.48 0.40 0.08 0.04

0.46 0.41 0.06 0.07

0.43 0.43 0.03 0.11

0.63 0.26 0.10 0.01

0.69 0.26 0.04 0.01

0.68 0.25 0.05 0.01

0.68 0.27 0.03 0.02

26 to 30 years Employed NILF Armed Forced Institutionalized

0.80 0.13 0.04 0.04

0.69 0.22 0.04 0.05

0.64 0.23 0.04 0.09

0.61 0.25 0.02 0.12

0.89 0.07 0.04 0.01

0.86 0.10 0.02 0.01

0.86 0.10 0.03 0.01

0.85 0.11 0.02 0.02

31 to 40 Employed NILF Armed Forced Institutionalized

0.82 0.11 0.04 0.03

0.76 0.18 0.03 0.03

0.70 0.21 0.03 0.06

0.64 0.23 0.02 0.11

0.91 0.05 0.03 0.01

0.90 0.07 0.02 0.01

0.89 0.08 0.02 0.01

0.87 0.10 0.01 0.02

41 to 50 Employed NILF Armed Forced Institutionalized

0.83 0.14 0.01 0.02

0.77 0.21 0.01 0.02

0.74 0.21 0.01 0.04

0.65 0.28 0.01 0.06

0.92 0.06 0.01 0.01

0.90 0.09 0.01 0.01

0.90 0.10 0.01 0.01

0.86 0.12 0.00 0.01

51 to 65 Employed NILF Armed Forced Institutionalized

0.72 0.26 0.00 0.02

0.61 0.37 0.00 0.01

0.58 0.40 0.00 0.02

0.53 0.44 0.00 0.03

0.81 0.18 0.00 0.01

0.72 0.27 0.00 0.01

0.69 0.31 0.00 0.01

0.70 0.29 0.00 0.01

Figures are tabulated from the 1970, 1980, 1990, and 2000 public use micro data samples from the U.S. Census of Population and Housing.

Table 3 Employment and Institutionalization Status for Non-Hispanic Black Males 40 and Under with a High School Education or Less , 1970 to 2000 High School Dropouts High School Graduates Age 1970 1980 1990 2000 1970 1980 1990 200018 to 25 years Employed NILF Armed Forced Institutionalized

0.50 0.38 0.04 0.08

0.38 0.51 0.04 0.08

0.30 0.55 0.00 0.15

0.27 0.50 0.00 0.23

0.62 0.23 0.13 0.02

0.52 0.32 0.13 0.03

0.49 0.36 0.10 0.06

0.44 0.44 0.04 0.09

26 to 30 years Employed NILF Armed Forced Institutionalized

0.76 0.16 0.01 0.06

0.58 0.32 0.01 0.10

0.40 0.38 0.00 0.22

0.30 0.36 0.00 0.34

0.83 0.09 0.06 0.02

0.70 0.21 0.05 0.04

0.64 0.24 0.04 0.08

0.58 0.29 0.02 0.12

31 to 40 Employed NILF Armed Forced Institutionalized

0.81 0.13 0.01 0.05

0.70 0.25 0.00 0.05

0.52 0.34 0.00 0.13

0.35 0.37 0.00 0.28

0.82 0.08 0.08 0.02

0.76 0.17 0.04 0.03

0.69 0.24 0.02 0.06

0.62 0.27 0.01 0.11

Figures are tabulated from the 1970, 1980, 1990, and 2000 public use micro data samples from the U.S. Census of Population and Housing.

Table 4 BJS Estimates of the Proportion of the Male Population Ever Having Served Time in a State or Federal Prison by Race/Ethnicity and Age and Estimates of the Proportion Serving Time in a California State Prison During the 1990s, by Race, Age and Educational Attainment Estimates for California from CDC Administrative Records

BJS estimates for the nationa

Allb

High school dropoutsc

High school graduatesc

Some collegec

College plusc

Non-Hispanic White Males 18 to 24 0.01 0.01 0.03 0.00 0.00 0.00 25 to 34 0.03 0.03 0.31 0.03 0.01 0.00 35 to 44 0.04 0.03 0.30 0.04 0.02 0.01 45 to 54 0.03 0.02 0.17 0.02 0.01 0.01 55 to 65 0.03 0.01 0.04 0.01 0.00 0.00 Non-Hispanic Black Males 18 to 24 0.09 0.04 0.19 0.02 0.01 0.00 25 to 34 0.20 0.19 1.14 0.15 0.05 0.03 35 to 44 0.22 0.19 1.23 0.16 0.07 0.04 45 to 54 0.18 0.15 0.90 0.12 0.06 0.05 55 to 65 0.13 0.05 0.18 0.04 0.01 0.02 Hispanic Males 18 to 24 0.04 0.01 0.02 0.00 0.00 0.00 25 to 34 0.09 0.05 0.08 0.03 0.02 0.02 35 to 44 0.10 0.05 0.07 0.04 0.02 0.03 45 to 54 0.10 0.03 0.04 0.03 0.02 0.03 55 to 65 0.07 0.01 0.02 0.02 0.01 0.01

a. Estimates drawn from Table 7 of Bonczar (2003) b. Estimates in this column are calculated as follows. The administrative term-records for all terms

served in California were sorted by a CDC internal id number. The first term for each unique id was selected out to construct a sample of unduplicated prisoners. For each prisoner, we calculate how old the prisoner would be in the year 2000. We then calculated counts of prisoners by age and race for 2000. Using the 2000 one percent PUMS, we then estimate the California population size for each age/race cell listed in the table. The figures in the table are the ratio of the prisoner counts to the 2000 census population estimate for each cell.

c. Estimates in this column are calculated as follows. We first calculate the counts of unduplicated prisoners by age and race following the procedures in note b. We then use data from the 1997 Survey of Inmates in State and Federal Corrections Facilities to estimate the educational attainment of prison inmates in the United States by race/ethnicity and age. We use these estimates to allocate the number of unduplicated prisoners within each age-race cell across the four educational groups (the CDC administrative data does not contain information on educational attainment). We then use the 2000 one percent PUMS to estimate the California population size of each age/race/education cell in the table. The figures in the table are the ratio of the prisoner counts hypothetically allocated across education groups to the 200 census population estimate for each cell.

Table 5 Comparisons of the White-Black Employment Rate Differentials, by Year and Level of Education Attainment, Using Alternative Base Populations to Calculate Employment Rates 1970 1980 1990 2000 High School Dropouts Non-Institutionalized Total Population

0.07 0.09

0.11 0.12

0.13 0.17

0.19 0.25

Difference 0.02 0.01 0.04 0.06 High School Non-Institutionalized Total Population

0.05 0.06

0.10 0.11

0.13 0.16

0.17 0.21

Difference 0.01 0.01 0.03 0.04 Some College Non-Institutionalized Total Population

0.04 0.04

0.08 0.10

0.08 0.10

0.10 0.13

Difference 0.00 0.02 0.02 0.03 College Graduates Non-Institutionalized Total Population

0.02 0.02

0.05 0.06

0.03 0.04

0.06 0.07

Difference 0.02 0.01 0.01 0.01 The figures in the table are the differences between the employment-to-population ratios for white males and black males by education group. In the rows labeled “Non-Institutionalized,” the non-institutionalized population is used as the base for calculating the underlying employment rates. In the rows labeled “Total Population,” the non-institutionalized plus the institutionalized are used as the base for calculating the underlying employment rates. These figures are tabulated from the 1970, 1980, 1990, and 2000 one percent PUMS files.

Table 6 Median Time Served (Years) in the California State Prison System by Term and by Race/Ethnicity for 1990 Prisoner Cohort, 18 to 25 Years of Age Term Served First Second Third Fourth Fifth or

higher All Inmates 1.02 0.68 0.62 0.53 0.49 White 0.94 0.62 0.58 0.53 0.48 Black 1.17 0.71 0.65 0.53 0.50 Hispanic 1.01 0.72 0.63 0.55 0.47 Tabulation are based on all individuals between the ages of 18 and 25 that entered the California state prison system during 1990 serving the first term of a commitment. The terms served column refer to the first and subsequent terms served by the 1990 cohort of inmates over the subsequent 10 years.

Table 7 Quartile Values of the Total Time Served During the 1990s and the Time Between the Date of First Admission and Date of Last Release for the 1990 Prison Cohort Between 18 and 25 Years of Age Panel A: Distribution of Total Time Served 25th Percentile 50th Percentile 75th Percentile All Inmates 1.44 2.79 4.81 White 1.43 3.09 5.12 Black 1.93 3.53 5.45 Hispanic 1.29 2.23 3.97 Panel B: Distribution of Time Between the Date of First Admission and the Date of Last Release 25th Percentile 50th Percentile 75th Percentile All Inmates 1.86 4.99 8.71 White 2.01 6.17 9.11 Black 2.88 6.42 9.16 Hispanic 1.44 3.65 7.62 Tabulation are based on all individuals between the ages of 18 and 25 that entered the California state prison system during 1990 serving the first term of a commitment. Tabulation of the percentiles of the two time distributions are based on all terms served over the subsequent 10 years.

Table 8 Regressions of the Proportion Employed Among the Non-Institutionalized on the Proportion Institutionalized (1) (2) (3) (4) (5) (6) Black

-0.045 (0.022)

-0.029 (0.021)

-0.073 (0.025)

-0.058 (0.013)

-0.038 (0.010)

-0.045 (0.015)

Black*1980

-0.073 (0.027)

-0.070 (0.027)

-0.010 (0.031)

-0.060 (0.016)

-0.057 (0.013)

-0.019 (0.019)

Black*1990

-0.085 (0.028)

-0.056 (0.027)

0.002 (0.032)

-0.069 (0.016)

-0.041 (0.013)

-0.008 (0.019)

Black*2000

-0.116 (0.027)

-0.064 (0.027)

-0.026 (0.030)

-0.097 (0.015)

-0.049 (0.013)

-0.056 (0.017)

Asian

-0.104 (0.054)

-0.099 (0.052)

-0.109 (0.051)

-0.096 (0.031)

-0.090 (0.025)

-0.092 (0.025)

Asian*1980

0.027 (0.064)

0.024 (0.061)

0.035 (0.060)

0.021 (0.036)

0.017 (0.029)

0.022 (0.029)

Asian*1990

0.038 (0.060)

0.035 (0.057)

0.045 (0.056)

0.027 (0.035)

0.023 (0.028)

0.025 (0.027)

Asian*2000

0.024 (0.0.57)

0.019 (0.055)

0.029 (0.054)

0.010 (0.033)

0.005 (0.027)

0.007 (0.025)

Hispanic

-0.025 (0.016)

-0.010 (0.016)

-0.048 (0.019)

-0.035 (0.009)

-0.017 (0.008)

-0.024 (0.012)

Hispanc* 1980

0.027 (0.027)

0.008 (0.026)

0.051 (0.028)

0.037 (0.016)

0.018 (0.013)

0.023 (0.015)

Hispanic* 1990

0.037 (0.024)

0.028 (0.023)

0.072 (0.026)

0.051 (0.014)

0.039 (0.011)

0.049 (0.015)

Hispanic* 2000

-0.035 (0.022)

-0.046 (0.020)

-0.007 (0.025)

-0.014 (0.013)

-0.035 (0.014)

-0.028 (0.014)

Institutionalized -

-1.007 (0.183)

0.938 (0.659)

- -1.076 (0.098)

-0.687 (0.602)

Institutionalized* 1980

- - -2.518 (0.710)

- - -1.851 (0.759)

Institutionalized* 1990

- - -2.220 (0.065)

- - -0.968 (0.639)

Institutionalized* 2000

- - -1.861 (0.637)

- - -0.167 (0.612)

Age/education dummies

Yes Yes Yes Yes Yes Yes

Age/education dummies*year

No No No Yes Yes Yes

R2 0.887 0.898 0.903 0.971 0.981 0.983 N 320 320 320 320 320 320 All models include a constant term and year dummy variables. The dependent variable is the proportion of the non-institutionalized age/race/education/year cell that is employed. The key explanatory variable is the proportion of each cell (including the institutionalized) that is institutionalized.